ABSTRACT. Theoretical predictions of red giant branch stars' effective temperatures, colors, luminosities, and surface chemical abundances are a necessary tool for the astrophysical interpretation of the visible-near-infrared integrated light from unresolved stellar populations, the color-magnitude diagrams of resolved stellar clusters and galaxies, and spectroscopic determinations of red giant chemical abundances. On the other hand, the comparison with empirical constraints provides a stringent test for the accuracy of present generations of red giant models.We review the current status of red giant star modeling, discussing in detail the still-existing uncertainties affecting the model input physics (e.g., electron conduction opacity, treatment of the superadiabatic convection) and the adequacy of the physical assumptions built into the model computations.We compare theory with several observational features of the red giant branch in Galactic globular clusters, such as the luminosity function "bump," the luminosity of the red giant branch tip, and the envelope chemical abundance patterns, to show the level of agreement between current stellar models and empirical data concerning the stellar luminosities, star counts, and surface chemical abundances.